I’m not really qualified to talk about this. But a friend of mine is a research physicist at Bristol in the UK.
Routine coding is part of his work like it is for most scientists these days, but he doesn’t think of himself as a coder.
He’s been doing some intriguing work over the past year with one of the LHC trigger detectors at CERN in Geneva, looking for ways to enhance and speed up particle detection. He’s using off-the-shelf machine-learning packages to increase efficiency and detection rates by something like an order of magnitude.
He’s going to publish in the next couple of months, and he expects that if his own code is not ultimately used at CERN, then at least his concepts will be. His results are that dramatic.
And this from a guy who is not a machine-learning engineer, and not even rightfully speaking a data scientist.